我使用 jackson 将 json 字符串映射到我的 HTModel 类，它基本上是一个简单的 Pojo。
class HTModel{}
public class Post extends HTModel {
public String id;
public String content;
public String author;
}
即使类嵌套在一起也能很好地工作。
public class Venue extends HTModel {
public ArrayList<Post> posts;
}
我设置了一个简单的 SqlLite 模式来根据模型的类型和 ID 缓存和索引这些模型。
我的问题是，如果模型包含其他模型的字段，我不想将 field 模型作为一个整体存储在数据库中。 ArrayList Venue.posts 中的每个帖子都应该单独保存。
最好的方法是什么？
最佳答案
在使用 JSON 创建自己的数据库 -> POJO 实现时，我遇到了类似的问题。这就是我解决问题的方法，对我来说效果很好。
让我们以您的 Post 对象为例。它需要很容易地表示为 JSON 对象并从 JSON 字符串创建。此外，它需要能够保存到数据库中。我已经根据这两个条件分解了我使用的类的继承关系:
Post
-> DatabaseObject
-> JsonObject
-> LinkedHashMap
从最基本的表示开始，一个JsonObject，它是一个扩展的LinkedHashMap。 Maps 由于其键值映射而非常适合表示 JSON 对象。这是 JsonObject 类:
import java.util.Iterator;
import java.util.LinkedHashMap;
import java.util.Map;
/**
* A <code>JsonObject</code> represents a JSON object, which begins and ends
* with curly braces '{' '}' and contains key-value pairs separated by a
* colon ':'.
* <p>
* In implementation, this is simply an extended <code>LinkedHashMap</code> to
* represent key-value pairs and to preserve insert order (which may be
* required by some JSON implementations, though is not a standard).
* <p>
* Additionally, calling <code>toString()</code> on the <code>JsonObject</code>
* will return a properly formatted <code>String</code> which can be posted as
* a value JSON HTTP request or response.
* @author Andrew
* @param <V> the value class to use. Note that all keys for a
* <code>JsonObject</code> are <code>Strings</code>
*/
public class JsonObject<V> extends LinkedHashMap<String, V> {
/**
* Creates a new empty <code>JsonObject</code>.
*/
public JsonObject() {
}
/**
* Creates a new <code>JsonObject</code> from the given HTTP response
* <code>String</code>.
* @param source HTTP response JSON object
* @throws IllegalArgumentException if the given <code>String</code> is not
* a JSON object, or if it is improperly formatted
* @see JsonParser#getJsonObject(java.lang.String)
*/
public JsonObject(String source) throws IllegalArgumentException {
this(JsonParser.getJsonObject(source));
}
/**
* Creates a new <code>JsonObject</code> from the given <code>Map</code>.
* @param map a <code>Map</code> of key-value pairs to create the
* <code>JsonObject</code> from
*/
public JsonObject(Map<? extends String, ? extends V> map) {
putAll(map);
}
/**
* Returns a JSON formatted <code>String</code> that properly represents
* this JSON object.
* <p>
* This <code>String</code> may be used in an HTTP request or response.
* @return JSON formatted JSON object <code>String</code>
*/
@Override
public String toString() {
StringBuilder sb = new StringBuilder("{");
Iterator<Map.Entry<String, V>> iter = entrySet().iterator();
while (iter.hasNext()) {
Map.Entry<String, V> entry = iter.next();
sb.append(JsonParser.toJson(entry.getKey()));
sb.append(':');
V value = entry.getValue();
sb.append(JsonParser.toJson(value));
if (iter.hasNext()) {
sb.append(',');
}
}
sb.append("}");
return sb.toString();
}
}
很简单，它只是一个代表JSON对象的LinkedHashMap，可以通过调用toString()快速将其转为JSON字符串，以及创建使用我创建的 JsonParser 类从 JSON 字符串中提取。
如果您已经在使用像 Jackson 这样的 JSON 解析器，您可能需要重新做一些事情才能使用该 API。
接下来是 Post 的内容，DatabaseObject 赋予 Post 与数据库通信的功能。在我的实现中，Database 对象只是一个抽象类。我指定 Database 如何将 DatabaseObjects 保存到别处，无论是通过 JDBC 还是 JSON over HTTP。
请记住，我们使用 Map 来表示我们的对象。对于您的帖子，这意味着您拥有三个“属性”(我在文档中称其为关键值):ID、内容和作者。
这是 DatabaseObject(精简后)的样子。请注意 save() 方法，我将在此处回答您的问题。
import java.text.ParseException;
import java.util.ArrayList;
import java.util.Iterator;
import java.util.LinkedHashMap;
import java.util.List;
import java.util.Map;
import java.util.Set;
/**
* The <code>DatabaseObject</code> represents a single row of data from a
* specific table within a database.
* <p>
* The object should implement getters and setters for each column, and is
* responsible for registering the correct table name and column names, as
* well as default values for those columns, in case a default value is
* not supported by the database table.
* <p>
* The <code>DatabaseObject</code> works with key-value pairs as an
* extended <code>LinkedHashMap</code>. It defines one property,
* <code>DatabaseObject.ROW_ID</code> which represents the unique
* identifier column for a table row. This column should always be an
* integer value. (Future implementations may allow for long values, but
* <code>Integer.MAX_VALUE</code> is well suited for most databases as a maximum
* row count per table).
* <p>
* The key and value pairs should be accessed by implementing getter and
* setter methods, not by the get and put methods provided by the
* <code>LinkedHashMap</code>. This is to ensure proper <code>Class</code>
* type casting for each value.
* <p>
* A <code>DatabaseObject</code> itself is also an extension of a
* <code>JsonObject</code>, and <code>toString()</code> may be called on
* it to provide a JSON notated <code>DatabaseObject</code>.
* <p>
* When using JSON however, keep in mind that the keys may not correspond
* exactly with the table column names, even though that is the recommendation.
* The <code>DatabaseObject</code> should be converted back into its
* implementing object form and saved when using web services.
* <p>
* The parameter <code>T</code> should be set to the class of the implementing
* <code>DatabaseObject</code>. This will allow proper class casting when
* returning instances of the implementation, such as in the <code>load()</code>
* methods.
* @param <T> the type of <code>DatabaseObject</code>
* @author Andrew
*/
public abstract class DatabaseObject<T extends DatabaseObject>
extends JsonObject<Object> implements Cloneable{
/**The property for the row ID*/
public final static String ROW_ID = "rowId";
/**
* Creates a new empty <code>DatabaseObject</code>.
*/
public DatabaseObject() {
}
/**
* {@inheritDoc }
* <p>
* This get method will additionally check the <code>Class</code> of
* the returned value and cast it if it is a <code>String</code> but
* matches another <code>Class</code> type such as a number.
* @see #doGet(java.lang.String, boolean)
*/
@Override
public Object get(Object key) {
//From here you can specify additional requirements before retrieving a value, such as class checking
//This is optional of course, and doGet() calls super.get()
return doGet(String.valueOf(key), true);
}
/**
* {@inheritDoc }
* <p>
* This get method will additionally check the <code>Class</code> of
* the given value and cast it if it is a <code>String</code> but
* matches another <code>Class</code> type such as a number.
* @see #doPut(java.lang.String, java.lang.Object, boolean)
*/
@Override
public Object put(String key, Object value) {
//Like doGet(), doPut() goes through additional checks before returning a value
return doPut(key, value, true);
}
//Here are some example getter/setter methods
//DatabaseObject provides an implementation for the row ID column by default
/**
* Retrieves the row ID of this <code>DatabaseObject</code>.
* <p>
* If the row ID could not be found, -1 will be returned. Note that
* a -1 <i>may</i> indicate a new <code>DatabaseObject</code>, but it
* does not always, since not all <code>Databases</code> support
* retrieving the last inserted ID.
* <p>
* While the column name might not correspond to "rowId", this
* method uses the <code>DatabaseObject.ROW_ID</code> property. All
* objects must have a unique identifier. The name of the column
* should be registered in the constructor of the object.
* @return the <code>DatabaseObject's</code> row ID, or -1 if it is not set
*/
public int getRowId() {
//getProperty(), while not included, simply returns a default specified value
//if a value has not been set
return getProperty(ROW_ID, -1);
}
/**
* Sets the row ID of this <code>DatabaseObject</code>.
* <p>
* <b>Note: this method should rarely be used in implementation!</b>
* <p>
* The <code>DatabaseObject</code> will automatically set its row ID when
* retrieving information from a <code>Database</code>. If the row ID
* is forcibly overriden, this could overwrite another existing row entry
* in the database table.
* @param rowId the row ID of the <code>DatabaseObject</code>
*/
public void setRowId(int rowId) {
//And again, setProperty() specifies some additional conditions before
//setting a key-value pair, but its not needed for this example
setProperty(ROW_ID, rowId);
}
//This is where your question will be answered!!
//Since everything in a DatabaseObject is set as a key-value pair in a
//Map, we don't have to use reflection to look up values for a
//specific object. We can just iterate over the key-value entries
public synchronized void save(Database database) throws SaveException {
StringBuilder sql = new StringBuilder();
//You would need to check how you want to save this, let's assume it's
//an UPDATE
sql.append("UPDATE ").append(getTableName()).append(" SET ");
Iterator<String, Object> iter = entrySet().iterator();
while (iter.hasNext()) {
Map.Entry<String, Object> entry = iter.next();
String property = entry.getKey();
Object value = entry.getValue();
if (value instanceof DatabaseObject) {
((DatabaseObject)value).save();
}
else if (value instanceof Collection) {
for (Object v : (Collection)value) {
((DatabaseObject)value).save();
}
}
//However many other checks you want, such as Maps, Arrays, etc
else {
sql.append(property); //Assuming the property equals the column name
sql.append("='").append(value).append("'");
}
if (iter.hasNext()) {
sql.append(", ");
}
}
//getIdColumn() would retrieve which column is being used as the identifier
sql.append("WHERE ").append(getIdColumn()).append("=").append(getRowId());
//Finally, take our SQL string and save the value to the Database
//For me, I could simply call database.update(sql), and
//the abstracted Database object would determine how to
//send that information via HTTP as a JSON object
//Of course, your save() method would vary greatly, but this is just a quick
//and dirty example of how you can iterate over a Map's
//key-value pairs and take into account values that are
//DatabaseObjects themselves that need to be saved, or
//a Collection of DatabaseObjects that need to be saved
}
/**
* Retrieves the name of the database table that this
* <code>DatabaseObject</code> pulls its information from.
* <p>
* It is recommended to declare a final and static class variable that
* signifies the table name to reduce resource usage.
* @return name of the database table
*/
public abstract String getTableName();
}
对于 TL;DR 版本:
Post 是一个 DatabaseObject。
DatabaseObject 是一个JsonObject，它是一个LinkedHashMap。
LinkedHashMap 设定了存储键值对的标准。键 = 列名，值 = 列值。
JsonObject 什么都不做，只是提供了一种将 LinkedHashMap 打印为 JSON 字符串的方法。
DatabaseObject 指定有关如何保存到数据库的逻辑，包括在值是另一个 DatabaseObject 的情况下，或者如果值包含 DatabaseObject ，例如集合。
^ -- 这意味着您只需编写一次“保存”逻辑。当您调用 Post.save() 时，它会保存帖子。当您调用 Venue.save() 时，它会保存地点列(如果有的话)，以及 ArrayList 中的每个单独的 Post。
为了额外的乐趣，下面是您的 Post 和 Venue 对象的样子:
public class Post extends DatabaseObject<Post> {
//The column names
public final static String POST_ID = "PostID";
public final static String CONTENT = "Content";
public final static String AUTHOR = "Author";
public Post() {
//Define default values
}
public int getPostId() {
return (Integer)get(POST_ID);
}
public void setPostId(int id) {
put(POST_ID, id);
}
public String getContent() {
return (String)get(CONTENT);
}
public void setContent(String content) {
put(CONTENT, content);
}
public String getAuthor() {
return (String)getAuthor();
}
public void setAuthor(String author) {
put(AUTHOR, author);
}
@Override
public String getTableName() {
return "myschema.posts";
}
}
请注意，我们不再声明类变量，只声明存储值的列名。我们在 getter/setter 方法中定义变量的类。
import java.util.ArrayList;
import java.util.List;
public class Venue extends DatabaseObject<Venue> {
//This is just a key property, not a column
public final static String POSTS = "Posts";
public Venue() {
setPosts(new ArrayList());
}
public List<Post> getPosts() {
return (List<Post>)get(POSTS);
}
public void setPosts(List<Post> posts) {
put(POSTS, posts);
}
public void addPost(Post post) {
getPosts().add(post);
}
@Override
public String getTableName() {
//Venue doesn't have a table name, it's just a container
//In your DatabaseObject.save() method, you can specify logic to
//account for this condition
return "";
}
}
额外的终极 TL;DR 版本:
使用 Map 来存储您的变量，而不是在您的类中定义它们。
创建一个 save() 方法逻辑，迭代 Map 值并将列值对保存到数据库，同时考虑集合或可保存的值数据库对象。
现在您所要做的就是调用 Venue.save()，您所有的 Post 对象也将被适本地保存。
关于java - 将嵌套的 Pojo 对象作为单独的对象存储在数据库中，我们在Stack Overflow上找到一个类似的问题： https://stackoverflow.com/questions/19726894/